initial file

app.py

@@ -0,0 +1,80 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
+import torch
+import streamlit as st
+from streamlit_chat import message
+
+checkpoint = "/Users/user/Documents/data/model/27-3-2023"
+#checkpoint = "microsoft/DialoGPT-large"
+
+@st.cache_resource
+def get_model():
+    model = AutoModelForCausalLM.from_pretrained(checkpoint)
+    tokenizer = AutoTokenizer.from_pretrained(checkpoint)
+    return model, tokenizer
+     
+st.title("Chat with myGPT 🦄")     
+st.write("This is a LLM that was fine-tuned on a dataset of daily conversations.")
+
+if 'count' not in st.session_state or st.session_state.count >= 3:
+    st.session_state.count = 0 
+    st.session_state.chat_history_ids = None
+    st.session_state.old_response = ''
+else:
+    st.session_state.count += 1
+    
+if 'message_history' not in st.session_state:
+    st.session_state.message_history = []
+
+if 'response_history' not in st.session_state:
+    st.session_state.response_history = []
+    
+if 'input' not in st.session_state:
+    st.session_state.input = ''
+
+def submit():
+    st.session_state.input = st.session_state.user_input
+    st.session_state.user_input = ''
+
+# prompt = "How long will it take for the poc to finish?"
+# inputs = tokenizer(prompt, return_tensors="pt")
+model, tokenizer = get_model()
+generation_config = GenerationConfig(max_new_tokens=32, 
+                        num_beams=4, 
+                        early_stopping=True,
+                        no_repeat_ngram_size=2, 
+                        do_sample=True, 
+                        penalty_alpha=0.6,
+                        top_k=4, 
+                        #top_p=0.95,
+                        #temperature=0.8,
+                        pad_token_id=tokenizer.eos_token_id)
+
+
+for i in range(0, len(st.session_state.message_history)):
+    message(st.session_state.message_history[i], is_user=True, key=str(i)+'_user', avatar_style="identicon", seed='You') # display all the previous message
+    if i in range(0, len(st.session_state.response_history)):
+        message(st.session_state.response_history[i], key=str(i), avatar_style="bottts", seed='mera GPT')
+        
+placeholder = st.empty() # placeholder for latest message
+st.text_input('You:', key='user_input', on_change=submit)
+
+if st.session_state.input:
+    st.session_state.message_history.append(st.session_state.input)
+    new_user_input_ids = tokenizer.encode(tokenizer.eos_token + st.session_state.input, return_tensors="pt")
+    bot_input_ids = torch.cat([st.session_state.chat_history_ids, new_user_input_ids], dim=-1) if st.session_state.count > 1 else new_user_input_ids
+
+    st.session_state.chat_history_ids = model.generate(bot_input_ids, generation_config)
+    response = tokenizer.decode(st.session_state.chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)
+
+    if st.session_state.old_response == response:
+        bot_input_ids = new_user_input_ids
+        st.session_state.chat_history_ids = model.generate(bot_input_ids, generation_config)
+        response = tokenizer.decode(st.session_state.chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)
+
+    #st.write(f"meraGPT: {response}")
+    st.session_state.old_response = response
+    st.session_state.response_history.append(response)
+    
+    with placeholder.container():
+        message(st.session_state.message_history[-1], is_user=True, key=str(-1)+'_user', avatar_style="identicon", seed='You') # display the latest message
+        message(st.session_state.response_history[-1], key=str(-1), avatar_style="bottts", seed='mera GPT') # display the latest message

requirements.txt

@@ -0,0 +1,68 @@
+aioice==0.7.6
+aiortc==1.4.0
+altair==4.2.2
+attrs==22.2.0
+av==10.0.0
+blinker==1.5
+cachetools==5.3.0
+certifi==2022.12.7
+cffi==1.15.1
+charset-normalizer==3.0.1
+click==8.1.3
+cryptography==39.0.1
+decorator==5.1.1
+dnspython==2.3.0
+entrypoints==0.4
+filelock==3.9.0
+gitdb==4.0.10
+GitPython==3.1.31
+google-crc32c==1.5.0
+huggingface-hub==0.12.0
+idna==3.4
+importlib-metadata==6.0.0
+Jinja2==3.1.2
+jsonschema==4.17.3
+markdown-it-py==2.1.0
+MarkupSafe==2.1.2
+mdurl==0.1.2
+netifaces==0.11.0
+numpy==1.24.2
+packaging==23.0
+pandas==1.5.3
+Pillow==9.4.0
+protobuf==3.20.3
+pyarrow==11.0.0
+pycparser==2.21
+pydeck==0.8.0
+pyee==9.0.4
+Pygments==2.14.0
+pylibsrtp==0.8.0
+Pympler==1.0.1
+pyOpenSSL==23.0.0
+pyrsistent==0.19.3
+python-dateutil==2.8.2
+pytz==2022.7.1
+pytz-deprecation-shim==0.1.0.post0
+PyYAML==6.0
+regex==2022.10.31
+requests==2.28.2
+rich==13.3.1
+semver==2.13.0
+six==1.16.0
+smmap==5.0.0
+streamlit==1.18.1
+streamlit-chat==0.0.2.1
+streamlit-webrtc==0.44.6
+tokenizers==0.13.2
+toml==0.10.2
+toolz==0.12.0
+torch==1.13.1
+tornado==6.2
+tqdm==4.64.1
+transformers @ git+https://github.com/huggingface/transformers.git@e3d832ff87c6ec997125deaa4f1b239db8f9e613
+typing_extensions==4.5.0
+tzdata==2022.7
+tzlocal==4.2
+urllib3==1.26.14
+validators==0.20.0
+zipp==3.13.0